APBRmetrics

Mathketball wrote:Basically what I am doing is taking specific stats that I feel say something about a given player. The theory behind which stats I choose has come kind of been a combination of what I feel, what I have read, and insight I have gained from personal contacts. I started these formulas a few years ago in college and got together with an assistant coach from my college's team about preparing a report for him. Then I altered my formulas slightly based on some of his insight.

Mathketball wrote:What I'm striving for is complete objectivity.

Not sure how what you wrote in the first quote can be reconciled with the statement in the second quote.

Mathketball wrote:Small Forwards:

**Carlton Scott: 1.5763
Derrick Williams: 1.5244
Kawhi Leonard: 1.4913
Tyler Honeycutt: 1.4624

* Guards tend to score higher then forwards and centers so to compare them straight across isn't quite fair. I'm working on scaling them to do away with that problem. Kevin Love was the exception to this trend.

** Obviously the clear question mark here is the remarkably high score by Carlton Scott. He is the one true enigma in this draft based on my scales. I can't really explain away his score but I certainly am not trying to claim he is the best SF in this draft. With where he is projected to go I think he is someone worth watching if he is able to stick in the NBA.

* Finally, I would say my true sleeper of the SF's is Honeycutt. In a draft that is so weak at SF to be able to get him late in the first could be a real nice steal for a team.

Well, I think I have a pretty good idea how you approach this - it appears you take players' stats in various categories and compare them to a mean, maybe weight the results for position (assists weighted more for PG, rebounds PF/C, etc), total, and then average by # of categories. I don't know how (if?) you incorporate pace or SoS. Anyway - it seems you've pretty much ranked players by a glorified versatility index. Players with more diverse stats (fill up the boxscore) will rate high - the Scottie Pippens of the world. I highlighted Scott & Honeycutt up there - two guys who don't score much but do a bit of everything statistically. Conley was like that in college - GREAT assist rate, assist/to ratio, steal rate. If you weigh those things the same as scoring & usage - he'll rate very high.

You will get some hits for sure - since more athletic college players tend to do more things (steals, blocks, etc) - and more athletic players tend to have a better shot for the NBA. I feel you may be undervaluing usage quite a bit here. Cory Joseph and his incredibly pedestrian stats (outside of a nice A/to ratio, 3pt %, & steal rate) being rted much higher than future NBA teammates Thompson & Hamilton just seems off. His being ahead of Kemba seems crazy.

I suggest you try your approach on current NBA players - do the final rankings make sense? I have a feeling you might see funky results there too - which might mean you may need to do some tweaking (incorporating usage more?). We don't want the Pippens of the world to rank in front of the MJs.

I'd really like to see these ratings for previous drafts. Statman, can you post your ratings for the 2008 draft, so we can see how they turned out compared to reality?

gfarkas wrote:

Mathketball wrote:Basically what I am doing is taking specific stats that I feel say something about a given player. The theory behind which stats I choose has come kind of been a combination of what I feel, what I have read, and insight I have gained from personal contacts. I started these formulas a few years ago in college and got together with an assistant coach from my college's team about preparing a report for him. Then I altered my formulas slightly based on some of his insight.

Mathketball wrote:What I'm striving for is complete objectivity.

Not sure how what you wrote in the first quote can be reconciled with the statement in the second quote.

Why? The whole idea is to remove biases by just looking at the numbers. I'm trying to build a formula based off of statistics that in theory say something about the players make-up or potential. I guess I don't understand how that wouldn't be objective.

Can you tell us what your ratings were for the 2008 draft?

Statman, thanks for your input. I think you are probably right that I need to look at usage more. I do incorporate SoS but do not incorporate pace. Does anyone know a site with college teams pace rankings? I've thought of trying to incorporate it in the past but I didn't know where to find the numbers.

Evanz, When I get more time I will put up some of my results from years past.

Mathketball wrote:Statman, thanks for your input. I think you are probably right that I need to look at usage more. I do incorporate SoS but do not incorporate pace. Does anyone know a site with college teams pace rankings? I've thought of trying to incorporate it in the past but I didn't know where to find the numbers.

Evanz, When I get more time I will put up some of my results from years past.

Draft Express has pace stats.

I assumed you were using kenpom. Isn't everything you want on there?

EvanZ wrote:I assumed you were using kenpom. Isn't everything you want on there?

KenPom has probably the best college stats.

Thanks a lot. I'm guessing "AdjTempo" is kenpom's equivalent to pace?

Here's my results from the 2008 draft:

Pg:
Derrick Rose: 1.6635
Russel Westbrook: 1.6002
D.J. Augustin: 1.5858
Jarred Bayless: 1.5453
George Hill: 1.4919

Sg:
O.J. Mayo: 1.5349
Brandon Rush: 1.4920
Eric Gordon: 1.4597
Sonny Weems: 1.4573
J.R. Giddens: 1.4466

* Gordon is clearly a bit lower than expected. With the turmoil around the team and the mid season coaching change that could possibly explain this.

Sf:
Micheal Beasley: 1.5961
Maarty Leunen: 1.5738
Donte Greene: 1.4561
Joe Alexander: 1.2974
Bill Walker: 1.1736

* There weren't very many SF's taken in this draft after Greene there is clearly a large drop off. Alexander scored very low to be picked at 8. The obvious case of a false positive here is Maarty Leunen.

Pf:
Kevin Love: 1.6479
Ryan Anderson: 1.3471
J.J. Hickson: 1.2875
James Gist: 1.2519
D.J. White: 1.2449

C:
Brook Lopez: 1.5183
Roy Hibbert: 1.4497
Robin Lopez: 1.3034
Marreese Speights: 1.2685
Kosta Kufos: 1.2417

* The guys I would say got false negative scores were JaVale McGee (C #8: 1.1181) and Deandre Jordan (C #9: 1.0549)

EvanZ wrote:I'd really like to see these ratings for previous drafts. Statman, can you post your ratings for the 2008 draft, so we can see how they turned out compared to reality?

Honestly, I could if I could get all the data compiled. I have the 2008 stats I believe (on a seperate hard drive somewhere - we stripped this computer clean recently) - but I don't think I have the players broken down by class. So, point is - I PLAN on getting this done - but it may take a little while considering I still work/family/exercise/sleep. I wish there was much more time in the day.

I'm pretty focused on the college stuff at the moment - and I have a stronger desire than ever to get all those past seasons compiled in a way so that I could do BOTH the normal ratings as well as rating potential stuff (now that I have a strong idea how I want to approach it), to see how the results look in retrospect.

EvanZ wrote:I'd really like to see these ratings for previous drafts. Statman, can you post your ratings for the 2008 draft, so we can see how they turned out compared to reality?

Hey Evan - I found the post a while back I made with the top 27 dudes in college basketball in 2008:

http://www.pointguardu.com/f136/statman ... post340490

Now - this IS NOT my rating potential stuff I've been doing the last few days - there are some highly ranked seniors (Aleks Maric!) that obviously wouldn't rank nearly as high when we looked at potential. My guess Beasley or possibly Love would be #1 on "potential" - followed by Curry, Rose, Harden, Mayo, Lopez, maybe Anderson, then maybe Hansbrough. Honestly - I have no clue how the order of those guys would end up, I think they'd be fairly closely bunched after Beasley & Love - it'd be great to find out.

I'm telling you - now that I figured out how I would like to approach it - I now REALLY need to work on getting all those past stats.

Very interesting. I definitely want to see how the potential rating turns out.

I figured you were using past season to develop your ratings, but I guess not?

Mathketball wrote:

gfarkas wrote:

Mathketball wrote:Basically what I am doing is taking specific stats that I feel say something about a given player. The theory behind which stats I choose has come kind of been a combination of what I feel, what I have read, and insight I have gained from personal contacts. I started these formulas a few years ago in college and got together with an assistant coach from my college's team about preparing a report for him. Then I altered my formulas slightly based on some of his insight.

Mathketball wrote:What I'm striving for is complete objectivity.

Not sure how what you wrote in the first quote can be reconciled with the statement in the second quote.

Why? The whole idea is to remove biases by just looking at the numbers. I'm trying to build a formula based off of statistics that in theory say something about the players make-up or potential. I guess I don't understand how that wouldn't be objective.

Okay. You write that basically the crux of your system is to use statistics that you "feel say something about a given player" and that these are based on "what I feel", and then are arbitrarily "altered" based on some "insight" you receive through "personal contacts".

Are the feelings that led you to select certain statistics over others reproducible? Can they be validated? Do you have empirical and measurable evidence to support their selection?

And even then, what about these alterations you subsequently make? Are these based on systematic observation, measurement, and experiment, or do you just play around with the numbers until they "feel" right? By making these alterations, could one argue you are, in fact, introducing biases (those of the assistant coach with whom you interacted), rather than removing biases?

I feel blocked shots are especially valuable. They don't just deny a basket, but there's an intimidation factor involved. The other team is going to be scared to try to get near the basket again after you block their shot. Clearly, a blocked shot objectively reduces the the other team's scoring by... say... 6 to 8 points. Therefore, logically, a blocked shot is worth around 7 times as much as every point scored. Objectively, you should make sure to include blocked shots in your system, and they should be weighted at least 7 times as much as points scored.

Also, dunks are clearly more valuable than other baskets. Again, it comes down to intimidation. When a guy dunks the ball, the other team knows he means business... it makes a strong statement about the player's make-up and potential. Based on my objective insight, I would say dunks are worth 3 times as much as other baskets, and a system such as yours should adjust accordingly.

I assume you'll incorporate these modifications into your system, and I look forward to seeing the results of the updated formula.